Hydraulic variability as the governing parameter in hidden hydropower assessment: A standardised framework for water infrastructures

Abstract

Water infrastructures host a largely unexploited hidden hydropower potential that could power distributed monitoring systems without external supply. However, its assessment remains challenging due to the lack of standardised, transferable evaluation procedures. This work develops a unified framework for hidden hydropower assessment, integrating multi-year operational velocity data with a physics-informed, neural-network-accelerated reduced-order model and an energy-based optimisation procedure that maximises annual harvested energy under realistic flow variability. The framework is validated across eighteen pilot sites spanning pressurised pipe networks, small open-channel systems and large open-channel environments. The cross-site analysis reveals two novel scientific findings. First, a statistically robust inverse relationship exists between the coefficient of variation of flow velocity (CV) and the capacity factor across all infrastructure categories ( R 2 = 0.8408 ), establishing CV as one of the dominant drivers of long-term energy utilisation that cannot be inferred from device-level, steady-flow studies. Second, the optimal design velocity may deviate from both the mean and the most frequent velocity, owing to the cubic scaling of extractable power with flow speed, confirming that variability-aware optimisation is physically necessary, not merely preferable. Together, these findings support a quantitative site suitability map classifying candidate locations as Suitable, Marginal or Unsuitable based solely on operational flow data, without requiring preliminary harvester sizing. Sites with CV < 0.1 and medium-to-high velocities achieve capacity factors above 80%, suitable for direct sensor powering, whereas sites with CV > 0.2 show severely limited utilisation regardless of instantaneous velocity.